A metric should answer a question.
Ask what you need to know: Was the campaign delivered? Did people engage? Did a supported and tracked business action occur? Those are different questions, and they need different measures.
Do not turn a proxy into a promise. An ad view is not a person, a click is not a customer, and a reported conversion is not automatically proof that an ad caused the outcome.
Three layers of measurement.
| Layer | Examples | What to clarify |
|---|---|---|
| Delivery | Ad views, confirmed placements, supported reach metrics. | Metric definition, measurement source, channel limits, and reporting period. |
| Engagement | Clicks, click-through rate, video completion, tagged-link activity. | Counting rules, denominator, device or channel context, and invalid-traffic handling. |
| Business response | Tracked leads, purchases, donations, or campaign-attributed value. | Event definition, permissions, tracking reliability, attribution window, and deduplication. |
Write down the assumptions.
Agree the source of truth, conversion definition, attribution approach, and reporting period before you compare results. Keep click-through and view-through measures distinct where they are reported.
Different platforms can count the same event differently. Document the differences rather than presenting inconsistent numbers as an exact match.
Useful calculations, clearly labeled.
- Click-through rate: clicks ÷ ad views × 100.
- Landing-page conversion rate: counted conversions ÷ the agreed visit or user denominator × 100.
- Return on ad spend: campaign-attributed revenue ÷ advertising spend. This is not profit or a guarantee of incremental revenue.
Use only metrics supported by the campaign setup. Unique reach, cross-device reporting, and conversion attribution are not interchangeable or universally available.